quark.torch.quantization.utils

Module Contents

Functions

quark.torch.quantization.utils.set_op_by_name(layer: Union[torch.nn.Module, torch.nn.ModuleList], name: str, new_module: torch.nn.Module) → None

Replaces a submodule in a given neural network layer with a new module(e.g. quantized module). The submodule to be replaced is identified by the ‘name’ parameter, which specifies the name of the submodule using dot notation. If the name includes dots, it navigates through nested submodules to find the specific layer to replace. Otherwise, it directly replaces the submodule in the provided layer.

Parameters: - layer: The top-level module containing the submodule. - name: name of the submodule, split by dots. - new_module: The new module to replace the existing one, for example the quantized module.

quark.torch.quantization.utils.t_exponent(t: torch.Tensor) → torch.Tensor

Get element exponents

Args:

t (torch.Tensor): Input tensor

Returns:

torch.Tensor: Exponents for each elements. NaN and Inf are treated as zeros.